2021.8.26.1129 - Various bugfixes for BEPUv2 upgrade
Hello everyone! Just a small patch, fixing up some bugs found with the BEPUv2 upgrade! I've got more coming, but I wanted to get those out first. If you find any more, please make a report on GitHub!
This build is compatible with previous one so no need to update, unless you want the bugfixes right away of course!
[h2]Optimizations:[/h2]
- Added some inlining compiler hints for vector and quaternion types for potential extra optimizations
[h2]Tweaks:[/h2]
- Meshes with invalid vertices (e.g. NaN or Infinity) will now have triangles/points affected by those filtered, rather than filtering out the entire mesh
-- This preserves collisions with the rest of the geometry that's still valid when there are a few stray invalid vertices
- Merged Dutch translation update by @AnotherFoxGuy
[h2]Bugfixes:[/h2]
- Fixed step up logic for CharacterController getting blocked by colliders with CharacterCollider set, but that do not actually cause collision response (e.g. Trigger or No Collision), resulting in it randomly breaking (reported by @Veer | CMO)
- Fixed step up logic not working when character collision is enabled for given locomotin module
- Fixed applying forces/impulses or other velocity changes to CharacterController that do not actually change the existing velocity causing the ground supports to be cleared, resultig in no traction (character being slippery) (reported by @Hayden, GH #2858)
- Fixed InstallLocomotionModule keeping the global transform of the installed module, rather than local one, causing it to be offset, rotated and scaled when installed on the user (reported by @Shifty | Quality Control Lead, GH #2856)
- Zero-width spaces (U+200B) are now respected when layout out text for rendering (reported by @Earthmark and @kazu0617 Neos:kazu)
- Fixed mesh colliders for extremelly large meshes (e.g. with vertices at 1e+38f offset) corrupting the physics simulation and resulting in a crash (reported by @orange and @kazu0617 Neos:kazu, GH #2855)
-- Those mesh colliders are now simply ignored and won't be part of the simulation
- Added workaround for some meshes producing invalid convex hulls (reported by @Kal)
- Invalid mesh vertices will no longer produce empty convex hull, but will simply be filtered out
This build is compatible with previous one so no need to update, unless you want the bugfixes right away of course!
[h2]Optimizations:[/h2]
- Added some inlining compiler hints for vector and quaternion types for potential extra optimizations
[h2]Tweaks:[/h2]
- Meshes with invalid vertices (e.g. NaN or Infinity) will now have triangles/points affected by those filtered, rather than filtering out the entire mesh
-- This preserves collisions with the rest of the geometry that's still valid when there are a few stray invalid vertices
- Merged Dutch translation update by @AnotherFoxGuy
[h2]Bugfixes:[/h2]
- Fixed step up logic for CharacterController getting blocked by colliders with CharacterCollider set, but that do not actually cause collision response (e.g. Trigger or No Collision), resulting in it randomly breaking (reported by @Veer | CMO)
- Fixed step up logic not working when character collision is enabled for given locomotin module
- Fixed applying forces/impulses or other velocity changes to CharacterController that do not actually change the existing velocity causing the ground supports to be cleared, resultig in no traction (character being slippery) (reported by @Hayden, GH #2858)
- Fixed InstallLocomotionModule keeping the global transform of the installed module, rather than local one, causing it to be offset, rotated and scaled when installed on the user (reported by @Shifty | Quality Control Lead, GH #2856)
- Zero-width spaces (U+200B) are now respected when layout out text for rendering (reported by @Earthmark and @kazu0617 Neos:kazu)
- Fixed mesh colliders for extremelly large meshes (e.g. with vertices at 1e+38f offset) corrupting the physics simulation and resulting in a crash (reported by @orange and @kazu0617 Neos:kazu, GH #2855)
-- Those mesh colliders are now simply ignored and won't be part of the simulation
- Added workaround for some meshes producing invalid convex hulls (reported by @Kal)
- Invalid mesh vertices will no longer produce empty convex hull, but will simply be filtered out